tDiaryの過去ログをJekyllに移行したい

require 'fileutils'
require 'hiki2md'
require 'yaml'

INPUT_DIR = 'diary_log'
OUTPUT_DIR = '_posts'

def parse_td2_file(path)
  entries = []
  current = nil
  skip_keys = ['Title:', 'Last-Modified:', 'Visible:', 'Format:']

  File.read(path).split("\n").each do |line|
    if line.strip == '.'
      entries << current if current
      current = nil
    elsif line.start_with?('Date:')
      current = { date: line.sub('Date: ', '').strip, lines: [] }
    elsif current
      unless skip_keys.any? { |k| line.start_with?(k) }
        current[:lines] << line
      end
    end
  end

  entries << current if current
  entries
end

def split_by_heading(lines)
  sections = []
  current = nil

  lines.each do |line|
    if line.strip.start_with?('!')
      # ! で始まる見出しを新規セクションとする
      if line.strip =~ /^![^!]/
        sections << current if current
        current = [line]
      else
        current << line if current
      end
    else
      current << line if current
    end
  end

  sections << current if current
  sections
end

def write_markdown_sections(entry)
  date_str = entry[:date]
  y = date_str[0..3]
  m = date_str[4..5]
  d = date_str[6..7]
  ymd = "#{y}-#{m}-#{d}"

  sections = split_by_heading(entry[:lines])

  sections.each_with_index do |section_lines, idx|
    next if section_lines.nil? || section_lines.empty?

    heading_line = section_lines.first
    title_text = heading_line.sub(/^!+/, '').strip

    content = section_lines.join("\n")
    content = Hiki2md.new.convert(content)
    markdown = convert_liquid_tags(content)

    front_matter_hash = {
      'title'      => title_text
    }
    front_matter = YAML.dump(front_matter_hash) + "---\n"

    output_dir = File.join(OUTPUT_DIR, y, m)
    FileUtils.mkdir_p(output_dir)
    filename = File.join(output_dir, "#{ymd}-#{idx + 1}.md")

    File.write(filename, front_matter + "\n" + markdown)
    puts "Wrote #{filename}"
  end
end

def convert_liquid_tags(text)
  # Jekyllでエラーが出ちゃったから、％は小文字に置き換えて
  text.gsub(/\{\{\s*(\w+.*?)\s*\}\}/, "{％ \1 ％}")
end

Dir.glob("#{INPUT_DIR}/**/*.td2") do |file|
  entries = parse_td2_file(file)
  entries.each { |entry| write_markdown_sections(entry) }
end

# _plugins/tdiary.rb
module Jekyll
  class IsbnTag < Liquid::Tag
    def initialize(tag_name, text, tokens)
      super
      @isbn = text.strip
    end

    def render(context)
      "<p>(ISBN: #{@isbn})</p>"
    end
  end
end

Liquid::Template.register_tag('isbn', Jekyll::IsbnTag)

# _plugins/render_title_generator.rb
class RenderTitleGenerator < Jekyll::Generator
  def generate(site)
    site.posts.docs.each do |post|
      title_line = post.content.lines.first.to_s.strip
      title_line.gsub!(/^#\s*/, "")
      title_line.gsub!(/\[([^\[\]]+?(?:\[[^\]]+\])?[^\]]*?)\]\((https?:\/\/[^\s)]+)\)/, '\1')

      context = Liquid::Context.new({ "page" => post.data }, {}, { site: site })
      rendered = Liquid::Template.parse(title_line).render!(context)
      post.data["title_rendered"] = rendered
    end
  end
end